What are your thoughts on whether to have virtual master and maybe media servers (if not using tapes) versus physical servers? Assuming both have equal resources available.
Main points seem to be:
DR recovery from ransomware and cybersecurity attacks is becoming more common it’s interesting that nearly all DR testing is for physical failure and backup systems are still implemented with that in mind. Not many DR tests and designs include logical (data) failure where you have to actually have to recover data and services rather simply fail-over, such as ransomware attack.
NBU fail-over on virtual is relatively simple and physical still manageable but which is best for VMware failure or ransomware where physical and NDMP backups still exist?
Each solution has it weak and strong points but you can always migrate V2P or P2V if your needs have changed. A lot of our clients use virtual master servers and it's a good starting point for the majority of small and medium installations. Virtual media server might work out for some clients but they need to consider a lot of performance related things.
Assuming performance not a issue.
The issue I see is unnecessary dependency of the virtual infrastructure. It could have a serious impact of recovery time. Its risk management: Run & maintain costs versus business costs of delayed service restoration. Some people may agree others disagree, then there are other factors.
I have always opted for the "backup direct to offsite location". Do not want to have original and backup data in same location - ever.
Meaning backup should end up in a location where none of the original data is located. That located should be heavily firewalled, and have "segregation of duties" implemented.
Usually meaning physical is needed.
Just my thoughts :)
My opinion for what it's worth.
A virtual master should be no issue - although this does rely on the virtual infrastructure being there, so unless you lose the entire site, it is unlikely that a failure would impact ALL your virtual infrastructure. If the storage/datastore is impacted, then there are bigger issues at stake and invoking your DR stratgey should get you up and running (quickly). A DR plan for any master server should be available (and tested), which with virtual does provide simpler solutions.
Even if performance is not an issue, I still prefer physical media servers. You have the ability to utilise off host backup for virtual machines (SAN transport), and backups don't have performance impacts on other VMs.
I've setup virtual masters servers and DR can be very efficient for a physical disaster, now looking I'm looking at the more common disaster these days - corporate ransomware & cyberattacks. These attacks can get replicated into real-time standby systems. Which results recovery from backups cause the standby systems are useless. Some places even have to do complete rebuilds cause they get reinfected from restores.
It doesn't matter whether the master is virtual or physical it has the same chance of being compromised. My theory is, being physical you recover yourself with having to worry about others, like virtual infrastructure.